Introduction

I created both songs on StableAudio.com Links to an external site. after trying numerous AI generators. I tested almost every option on the Fairly Trained site, but only StableAudio.com Links to an external site. worked well for me. To make the songs, I used various keywords to create a chill guitar vibe. Both are instrumental since I find AI vocals too robotic and can’t sing myself, so I left out vocals. For song one, I used keywords like Post-Rock, Euphoric, and Sentimental (125 BPM). The second song features a 3/4 beat with Bright, Happy, and Surf vibes. Overall, I’m happy with the results and how natural they sound.

Visualisation

track 1

  • Approachability: 0.493762344121933
  • Arousal: 4.817852020263672
  • Danceability: 0.1733139157295227
  • Engagingness: 0.6132141947746277
  • Instrumentalness: 0.5142947435379028
  • Tempo: 86
  • Valence: 4.824087142944336

track 2

  • Approachability: 0.8854424357414246
  • Arousal: 4.87969970703125
  • Danceability: 0.28600555658340454
  • Engagingness: 0.5910114645957947
  • Instrumentalness: 0.349807471036911
  • Tempo: 83
  • Valence: 5.514646530151367

“features/marit-r-1.json” |>
compmus_chroma(norm = “Manhattan”) |>
ggplot(aes(x = time, y = pc, fill = value)) + geom_raster() + scale_y_continuous( breaks = 0:11, minor_breaks = NULL, labels = c( “C”, “C#|Db”, “D”, “D#|Eb”, “E”, “F”, “F#|Gb”, “G”, “G#|Ab”, “A”, “A#|Bb”, “B” ) ) + scale_fill_viridis_c(guide = “none”) +
labs(x = “Time (s)”, y = NULL, fill = NULL) + theme_classic()